[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Combat Simulations and Swarm
From: |
Rob Kewley |
Subject: |
Combat Simulations and Swarm |
Date: |
Fri, 27 Jun 1997 14:53:40 -0300 |
To Lance Forbes and Mike Brown:
I am answering you both over the swarm-support list because I wondered
if any other swarm users out there were working on similar problems.
I'll send most future messages directly so as not to clog the
swarm-support list.
As to your question on combat models in Swarm. I am currently working
on a high resolution combat model designed to simulate the battalion
and below mounted battle. I am not creating this model for the sake of
the model alone. My eventual goal is to parameterize the inputs to the
simulation that describe a friendly course of action. Then I will run
the simulation alongside a genetic algorithm to optimize the outcome of
the battle as a function of those course of action inputs. I have
already done this using a very simplistic model written in C. See my
web page at http://www.rpi.edu/kewler/ for a more detailed description
of this work. I believe that this model will have an application as a
tactical decision aid.
My current status is that I have a fairly simple model working in
swarm. It simulates the battle with one homogeneous force against
another. It fights on a terrain data base depicting a 10km by 10km
piece of terrain at the National Training Center. It displays the
terrain (I only use elevation data currently) and the friendly and enemy
forces as they fight the battle. The combat algorithms use a constant
rate movement model, a simplistic target acquisition model based on the
DYNTACS model (It considers line of sight as a regression model to fit
field data. Input variables are range, enemy height, enemy speed,
terrian complexity, and probability of looking in the sector containing
the enemy), a random target selection model, and a simple attrition
model which bases probability of kill on a linear interpolation between
100% at zero range and 0% at max range.
My strategy was to keep the model simple until I could build the
graphics, genetic algorithm, and fuzzy fitness function into it. Then I
will build in more complexity and realism. My preliminary work had
promising results.
My biggest problem is access to currently used combat models. I have
material from the Naval Postgraduate School, the ELAN combat model from
TRADOC Analysis Center - White Sands Missile Range, and the Military OR
Analyst's Handbook from Military Operation Research Society. I
currently only have access to terrain data on the US Geologic Survey
web site. If any of you have access to combat algorithms or processed
terrain data bases designed for simulations, I would be interested.
If you have any further questions, please email me directly. I will
send each of you periodic updates as long as you are interested.
==================================
Swarm-Support is for discussion of the technical details of the day
to day usage of Swarm. For list administration needs (esp.
[un]subscribing), please send a message to <address@hidden>
with "help" in the body of the message.
==================================
- Combat Simulations and Swarm,
Rob Kewley <=